Re: [Lsr] AD Review of draft-ietf-lsr-isis-srv6-extensions-11

["Les Ginsberg (ginsberg)" <[e-mail]>]

Alvaro/Peter -

In regards to:

> ...
> > 906 12.5. Sub-Sub-TLVs for SID Sub-TLVs
> >
> > 908 This document requests a new IANA registry be created under the IS-IS
> > 909 TLV Codepoints Registry to control the assignment of sub-TLV types
> > 910 for the SID Sub-TLVs specified in this document - Section 7.2,
> > 911 Section 8.1, Section 8.2. The suggested name of the new registry is
> > 912 "Sub-Sub-TLVs for SID Sub-TLVs". The registration procedure is
> > 913 "Expert Review" as defined in [RFC8126]. The following assignments
> > 914 are made by this document:
> >
> > [minor] In line with the name of other registries; suggestion:
> > "Sub-sub-TLVs for sub-TLVs 5, 43 and 44 (SRv6 End SID , SRv6 End.X SID
> > and SRv6 LAN End.X SID)".
> >
> >
> > ##PP2
> > I find that confusing as the sub-TLVs 5 is a locator Sub-TLV, while
> > Sub-TLVs 43 and 44 are IS reachability sub-TLVs.
> > What about:
> >
> > "Sub-Sub-TLVs for SID Sub-TLVs (SRv6 End SID, SRv6 End.X SID
> > and SRv6 LAN End.X SID)"
> 
> Most of the other registries include the number of the TLV.  So I
> think it would be good to remain consistent.  Maybe we should ask the
> current DEs: Chris, Hannes and Les.
> 

I think this is an odd situation.
You have three new sub-TLVs:

SRv6 End SID sub-TLV(5) which is a sub-TLV of TLVs 27, 135, 235, 236 and 237 (only allowed in TLV 27)

SRv6 End.X SID(43) and SRv6 LAN End.X SID(44) sub-TLVs which are sub-TLVs of TLVs 22, 23, 25, 141, 222 and 223 (allowed in all TLVs)

You propose to create a single registry common to all three of these sub-TLVs to share codepoints for sub-sub-TLVs. This is motivated by the one defined sub-sub-TLV SRv6 SID Structure Sub-Sub-TLV(1) which is applicable to all three new sub-TLVs.

Such a registry  (code points for sub-sub-TLVs associated with sub-TLVs in different parent TLVs) has not been defined before. Closest we have come is the "Sub-sub-TLV Codepoints for Application-Specific Link Attributes" which exists separately from " sub-TLVs of TLVs 22, 23, 25, 141, 222 and 223" but comes with the advisory:

"If a link attribute can be advertised both as a sub-TLV of TLVs
22, 23, 25, 141, 222, and 223 and as a sub-sub-TLV of the 
Application-Specific Link Attributes sub-TLV defined in 
[RFC8919], then the same numerical code
should be assigned to the link attribute whenever possible."

You could elect to do the same thing here i.e., create two new registries but include a Note saying codepoints in the two should be the same whenever it makes sense to do so.
Or, you could do something new and create a single registry even though the grandparent TLVs are different.

I don’t have a strong opinion here, but if it is anticipated that most of the sub-sub-TLVs would be shared then a single registry is easier to manage.

The name however becomes quite a mouthful - something like:

"sub-sub-TLVs for SRv6EndSID(5) (sub-TLV of TLVs 27, 135, 235, 236 and 237) and SRv6 End.X SID(43)/SRv6 LAN End.X SID(44) (sub-TLVs of TLVs 27, 135, 235, 236 and 237)"

And you would need to create columns (analogous to the columns in existing registries which have multiple parent TLVs) to show in which parents a given sub-sub-TLV is allowed.

   Les